[Profiler] Fix ASAN Overflow Issues #140441
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/140441
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 85c0a65 with merge base 3d2dd14 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D65835650 |
sraikund16
force-pushed
the
export-D65835650
branch
from
27a6b19 to
f71bb59
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D65835650 |
sraikund16
added
ciflow/trunk
Summary: It seems like this issues is due to leftover cupti events during warmup staying persistent in the queue during profiling. These events start before our actual time window and therefore have a timestamp lower than our basetime. This makes the delta become negative which results in unsigned overflow. This then creates a large number which later gets sign added which creates the signed overflow. Solution: If a raw timestamp is less than the base timestamp, just mark the process timestamp as -1 so we can mark these events as "to ignore". In Kineto, these events will be filtered as they will be out of range Test Plan: Tested with ASAN and was able to get a reasonable profiler. Also tested basic resnet test and only events in range show up Differential Revision: D65835650
sraikund16
force-pushed
the
export-D65835650
branch
from
f71bb59 to
85c0a65
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D65835650 |
davidberard98
approved these changes
Nov 13, 2024
| return [=](approx_time_t t_approx) { | ||
| // See above for why this is more stable than `A * t_approx + B`. | ||
| return (time_t)((double)(t_approx - t0_approx) * scale_factor) + t0; | ||
| return t_approx > t0_approx |
Contributor
There was a problem hiding this comment.
you could probably do
(time_t)(((double) t_approx - t0_approx) * scale_factor) + t0
without really affecting the precision, since in the original code you're already casting to double.
But if you prefer the current approach that's probably fine too
Contributor
|
@pytorchbot merge (Initiating merge automatically since Phabricator Diff has merged) |
Collaborator
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
pobin6
pushed a commit
to pobin6/pytorch
that referenced
this pull request
Dec 5, 2024
Summary: It seems like this issues is due to leftover cupti events during warmup staying persistent in the queue during profiling. These events start before our actual time window and therefore have a timestamp lower than our basetime. This makes the delta become negative which results in unsigned overflow. This then creates a large number which later gets sign added which creates the signed overflow. Solution: If a raw timestamp is less than the base timestamp, just mark the process timestamp as -1 so we can mark these events as "to ignore". In Kineto, add a special case to ignore timestamps that are negative. Test Plan: Test with ASAN Differential Revision: D65835650 Pull Request resolved: pytorch#140441 Approved by: https://github.com/davidberard98
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
It seems like this issues is due to leftover cupti events during warmup staying persistent in the queue during profiling. These events start before our actual time window and therefore have a timestamp lower than our basetime. This makes the delta become negative which results in unsigned overflow. This then creates a large number which later gets sign added which creates the signed overflow.
Solution: If a raw timestamp is less than the base timestamp, just mark the process timestamp as -1 so we can mark these events as "to ignore". In Kineto, add a special case to ignore timestamps that are negative.
Test Plan: Test with ASAN
Differential Revision: D65835650
cc @robieta @chaekit @guotuofeng @guyang3532 @dzhulgakov @davidberard98 @briancoutinho @sanrise